The blind source separation problem has received a great deal of attention in previous years. The aim of this problem is to estimate a set of original source signals from a set of linearly mixed signals through any number of signal processing techniques. While many methods exist that attempt to solve the blind source separation problem, a new technique is being used that uniquely separates audio sources as they are received from a microphone array. In this thesis a new algorithm is proposed that that utilizes the ICA algorithm in conjunction with a filtering technique that separates source signals and then removes sources of interference so that a signal of interest can be accurately tracked. Experimental results will compare a common blind source separation technique to the new algorithm and show that the new algorithm can detect a signal of interest and accurately track it as it moves through an anechoic environment. / Master of Science

This thesis describes the design and implementation of a 4-channel microphone array that is an adaptive beamformer used for hands-free telephony in a noisy environment. The microphone signals are amplified, then sent to an A/D converter. The microprocessor board takes the data from the 4 channels and utilizes digital signal processing to determine the direction-of-arrival of the sources and create an output which 'steers' the microphone array to the desired look direction while trying to minimize the energy of interference sources and noise. All of the processing for this thesis will be done on a computer using MATLAB. The MUSIC algorithm is used for direction finding in this thesis. It is shown to be effective in estimating direction-of-arrival for 1 speech source and 2 speech sources that are spaced fairly apart, with significant results down to a -5 dB SNR even. The MUSIC algorithm requires knowledge of the number of sources a priori, requiring an estimator for the number of sources. Though proposed estimators for the number of sources were examined, an effective estimator was not encountered for the case where there are multiple speech sources. Beamforming methods are examined which utilize knowledge of the source direction-of-arrival from the MUSIC algorithm. The input is split into 6 subbands such that phase-steered beamforming would be possible. Two methods of phase-steered beamforming are compared in both narrowband and wideband scenarios, and it is shown that phase-steering the array to the desired source direction-of-arrival has about 0.3 dB better beamforming performance than the simple time-delay steered beamformer using no subbands. As the beamforming solution is inadequate to achieve desired results, a generalized sidelobe canceler (GSC) is developed which incorporates a beamformer. The sidelobe canceler is evaluated using both NLMS and RLS adaptation. The RLS algorithm inherently gives better results than the NLMS algorithm, though the computational complexity renders the solution impractical for implementation with today's technology. A testing setup is presented which involves a linear 4-microphone array connected to a DSP chip that collects the data. Tests were done using 1 speech source and a model of the car noise environment. The sidelobe canceler's performance using 6 subbands (phase-delay GSC) and using 1 band (time-delay GSC) with NLMS updating are compared. The overall SNR improvement is determined from the signal and noise input and output powers, with signal-only as the input and noise-only as the input to the GSC. The phase-delay GSC gives on average 7.4 dB SNR improvement while the time-delay GSC gives on average 6.2 dB SNR improvement. / Master of Science

The speech of a person speaking in a noisy environment can be enhanced through electronic beamforming using spatially distributed microphones. As this approach demands precise information about the microphone locations, its application is limited in places where microphones must be placed quickly or changed on a regular basis. Highly precise calibration or measurement process can be tedious and time consuming. In order to understand tolerable limits on the calibration process, the impact of microphone position error on the intelligibility is examined. Analytical expressions are derived by modeling the microphone position errors as a zero mean uniform distribution. Experiments and simulations were performed to show relationships between precision of the microphone location measurement and loss in intelligibility. A variety of microphone array configurations and distracting sources (other interfering speech and white noise) are considered. For speech near the threshold of intelligibility, the results show that microphone position errors with standard deviations less than 1.5cm can limit losses in intelligibility to within 10% of the maximum (perfect microphone placement) for all the microphone distributions examined. Of different array distributions experimented, the linear array tends to be more vulnerable whereas the non-uniform 3D array showed a robust performance to positional errors.

Dans ce travail nous étudions la problématique des ingénieurs du son face au mixage d’un microphone principal HOA avec des microphones d’appoint, et notamment l’estimation des paramètres tels que le retard, la position et le gain des sources acoustiques associées aux microphones d’appoint. Nous proposons un algorithme fournissant les paramètres estimés (retard, position, gain) basé sur des équations d’encodage spatial au format HOA qui peuvent alors être utilisées pour traiter les signaux des microphones d’appoint durant le mixage. Cette extraction automatique des paramètres peut être vue comme une assistance pour les ingénieurs du son, leur permettant d’éviter un travail à faible valeur ajoutée (mesure de la distance et des angles entre microphones) afin de pouvoir se concentrer sur des problèmes artistiques comme l’ajustement des paramètres de niveau, d’égalisation ou de compression, voire l’ajustement fin des paramètres de retard, position, gain. La robustesse de l’algorithme est bien présentée pour les scènes sonores de différents niveaux de complexité (plusieurs sources acoustiques, réverbération, encodage réel du microphone…). Nous proposons des tests de performances pour les scènes sonores simulées et réels afin de montrer l’efficacité de l’algorithme ainsi que ces limites. La conclusion et les perspectives pour des futurs travaux complètent cette thèse à la fin du document. / In this thesis we study the problematic of a sound engineer mixing HOA (Higher Order Ambisonics) and spot microphones, namely the estimation of parameters such as delay, position and gain of acoustic sources associated to spot microphones. We present a typical workflow in this context, and also propose an algorithm extracting parameters that could be applied to the spot microphone signals. This mixing assistance allows sound engineers to easily work with HOA 3D sound and to concentrate on artistic choices (fine adjustments of the parameters), by avoiding a low-added value work (coarse parameter estimation). The robustness of the estimators is evaluated on recorded and artificial sound scenes, with different degrees of complexity in terms of number of sources and acoustic conditions (reverberation, effect of real microphone encoding, …). We also provide performance evaluations, based on both sound scene simulations and real recordings, showing encouraging results along with actual limits, and conclude on perspectives.

Dans ce travail, une approche de reconnaissance de l’orateur en utilisant un microphone de contact est développée et présentée. L'élément passif de contact est construit à partir d'un matériau piézoélectrique. La position du transducteur piézoélectrique sur le cou de l'individu peut affecter grandement la qualité du signal recueilli et par conséquent les informations qui en sont extraites. Ainsi, le milieu multicouche dans lequel les vibrations des cordes vocales se propagent avant d'être détectées par le transducteur est modélisé. Le meilleur emplacement sur le cou de l’individu pour attacher un élément transducteur particulier est déterminé en mettant en œuvre des techniques de simulation Monte Carlo et, par conséquent, les résultats de la simulation sont vérifiés en utilisant des expériences réelles. La reconnaissance est basée sur le signal généré par les vibrations des cordes vocales lorsqu'un individu parle et non sur le signal vocal à la sortie des lèvres qui est influencé par les résonances dans le conduit vocal. Par conséquent, en raison de la nature variable du signal recueilli, l'analyse a été effectuée en appliquant la technique de transformation de Fourier à court terme pour décomposer le signal en ses composantes de fréquence. Ces fréquences représentent les vibrations des cordes vocales (50-1000 Hz). Les caractéristiques en termes d'intervalle de fréquences sont extraites du spectrogramme résultant. Ensuite, un vecteur 1-D est formé à des fins d'identification. L'identification de l’orateur est effectuée en utilisant deux critères d'évaluation qui sont la mesure de la similarité de corrélation et l'analyse en composantes principales (ACP) en conjonction avec la distance euclidienne. Les résultats montrent qu'un pourcentage élevé de reconnaissance est atteint et que la performance est bien meilleure que de nombreuses techniques existantes dans la littérature. / In this work, a speaker recognition approach using a contact microphone is developed and presented. The contact passive element is constructed from a piezoelectric material. In this context, the position of the piezoelectric transducer on the individual’s neck may greatly affect the quality of the collected signal and consequently the information extracted from it. Thus, the multilayered medium in which the sound propagates before being detected by the transducer is modeled. The best location on the individual’ neck to place a particular transducer element is determined by implementing Monte Carlo simulation techniques and consequently, the simulation results are verified using real experiments. The recognition is based on the signal generated from the vocal cords’ vibrations when an individual is speaking and not on the vocal signal at the output of the lips that is influenced by the resonances in the vocal tract. Therefore, due to the varying nature of the collected signal, the analysis was performed by applying the Short Term Fourier Transform technique to decompose the signal into its frequency components. These frequencies represent the vocal folds’ vibrations (50-1000 Hz). The features in terms of frequencies’ interval are extracted from the resulting spectrogram. Then, a 1-D vector is formed for identification purposes. The identification of the speaker is performed using two evaluation criteria, namely, the correlation similarity measure and the Principal Component Analysis (PCA) in conjunction with the Euclidean distance. The results show that a high percentage of recognition is achieved and the performance is much better than many existing techniques in the literature.

Die Messung der Schichtdicke und die Charakterisierung von Oberflächen ist in modernen Produktionsprozessen von elementarem Interesse. Fehler wirken sich nicht nur unmittelbar sondern auch in der Weiterverarbeitung zum Endprodukt aus. In der Praxis werden zunehmend immer dünnere Schichten und edlere Materialien verarbeitet, was neue Verfahren notwendig macht. In dieser Arbeit sind mechanischer Aufbau, optische Komponenten, elektronische Bauteile und Eigenschaften eines Messsystems beschrieben, welches erstmals in der Lage ist, den Mengenauftrag trockener Konversions-Schutzschichten auf Aluminiumsubstraten hochempfindlich, berührungslos und nicht-destruktiv zu bestimmen. Das Messsystem ist bereits seit einigen Monaten im industriellen Einsatz. Die Messergebnisse wurden im Rahmen dieser Arbeit in einen Kontext vielfältiger Analysemethoden gesetzt und es konnte gezeigt werden, dass die anorganischen Komponenten der Konversionsschicht maßgeblichen Anteil an der Signalbildung haben. Entscheidend für die Qualität der Beschichtung von Substraten mit Vorbehandlung ist die Bestimmung des Zirkoniumgehalts. Die Robustheit der Methode gegenüber Störungen durch andere Parameter (Abstand, Energie, Trocknungs- Temperatur und anderen) wurde erwiesen. Im Vergleich mit anderen kommerziell verfügbaren Methoden zeigt das entwickelte Messsystem keine größeren Schwächen. Es kann prinzipiell nicht nur für Konversionsschichten, sondern außerdem für viele andere Schichtsysteme genutzt werden. Das in dieser vorgestellte Messsystem und seine grundlegende Funktionsweise wurden am 18. Februar 2014 zum Patent in den USA eingereicht.

L’aéroacoustique est une filière de l'acoustique qui étudie la génération de bruit par un mouvement fluidique turbulent ou par les forces aérodynamiques qui interagissent avec les surfaces. Ce secteur en pleine croissance a attiré des intérêts récents en raison de l’évolution de la transportation aérienne, terrestre et spatiale. Les microphones avec une bande passante de plusieurs centaines de kHz et une plage dynamique couvrant de 40Pa à 4 kPa sont nécessaires pour les mesures aéroacoustiques. Dans cette thèse, deux microphones MEMS de type piézorésistif à base de silicium polycristallin (poly-Si) latéralement cristallisé par l’induction métallique (MILC) sont conçus et fabriqués en utilisant respectivement les techniques de microfabrication de surface et de volume. Ces microphones sont calibrés à l'aide d'une source d’onde de choc (N-wave) générée par une étincelle électrique. Pour l'échantillon fabriqué par le micro-usinage de surface, la sensibilité statique mesurée est 0.4μV/V/Pa, la sensibilité dynamique est 0.033μV/V/Pa et la plage fréquentielle couvre à partir de 100 kHz avec une fréquence du premier mode de résonance à 400kHz. Pour l'échantillon fabriqué par le micro-usinage de volume, la sensibilité statique mesurée est 0.28μV/V/Pa, la sensibilité dynamique est 0.33μV/V/Pa et la plage fréquentielle couvre à partir de 6 kHz avec une fréquence du premier mode de résonance à 715kHz. / Aero-acoustics, a branch of acoustics which studies noise generation via either turbulent fluid motion or aerodynamic forces interacting with surfaces, is a growing area and has received fresh emphasis due to advances in air, ground and space transportation. Microphones with a bandwidth of several hundreds of kHz and a dynamic range covering 40Pa to 4kPa are needed for aero-acoustic measurements. In this thesis, two metal-induced-lateral-crystallized (MILC) polycrystalline silicon (poly-Si) based piezoresistive type MEMS microphones are designed and fabricated using surface micromachining and bulk micromachining techniques, respectively. These microphones are calibrated using an electrical spark generated shockwave (N-wave) source. For the surface micromachined sample, the measured static sensitivity is 0.4μV/V/Pa, dynamic sensitivity is 0.033μV/V/Pa and the frequency range starts from 100kHz with a first mode resonant frequency of 400kHz. For the bulk micromachined sample, the measured static sensitivity is 0.28μV/V/Pa, dynamic sensitivity is 0.33μV/V/Pa and the frequency range starts from 6kHz with a first mode resonant frequency of 715kHz.

In the late 1970s, a method was developed to estimate acoustic intensity in one dimension by taking the cross-spectral density of two closely-spaced microphone signals. Since then, multimicrophone probes have been developed to measure three-dimensional intensity as well as energy density. Their usefulness has led to the design of various types of multimicrophone probes, the most common being the four-microphone orthogonal, the four-microphone regular tetrahedron, and the six-microphone designs. These designs generally either consist of microphones suspended in space near each other or mounted on the surface of a sphere. This work analytically compares the relative merits of each probe design in measuring acoustic intensity and investigates the various finite-sum and finite-difference processing methods used with each. The analysis is limited to probes consisting of perfect point sensors in plane wave fields. The comparison is given in terms of average and maximum errors for intensity magnitude and direction as a function of angle of incidence as well as the spread between maximum and minimum errors for intensity magnitude. After existent probe geometries are reviewed, optimization techniques are introduced to predict what the optimal probe geometry would be for any given scenario. The probe is optimized to give the lowest intensity error averaged over angle of incidence of plane waves. This is done for full-space and half-space scenarios.

Research on microphone choice and placement has been previously made for a variety of instruments, such as drums, vocals, and guitar, to name a few. However, in comparison, very little research has been made on the bass-cabinet. With help of the different methodologies used for previous research on electric guitar and snare drum, the optimal placement and choice of microphone for a 4x10” bass-cabinet were investigated. In a listening test, pre-recorded basslines were mixed with other instruments since it is more common to hear it in a mix of instruments rather than listening to it by itself. To make sure that the topic is not too broad, focus on the genre of rock was taken for this investigation. Both qualitative and quantitative data were collected in a set of listening tests where the subjects rated their preference for their most and least preferable basslines. The subjects were asked to describe their preference utilizing different characteristics of the stimuli. An ANOVA test provided evidence that there is a statistical difference between the preference of one microphone model at a 15 cm distance from the bass-cabinet, in the category bass with band.

Typescript (photocopy). / Digitized by Kansas Correctional Industries